There is a pressing need for a workforce with the modeling and simulation skills associated with computational science. A number of national studies have substantiated those needs with respect to the future competitiveness of USA in research and development, the innovation of new products, and the competitiveness of many industries. Creating computational science programs at academic institutions organized along disciplinary boundaries represents a challenge because aspects of computer science, mathematics, and a science or engineering domain are required parts of any program. Gaining agreement on the associated requirements, integrating the classes with those in the traditional curriculum, and obtaining the necessary support through academic and administrative reviews represent substantial challenges. Clark Atlanta University, the University of Mary Washington, and Southern University are all working on establishing undergraduate or graduate programs in computational science and have had a number of common experiences. Copyright © 2014 John Wiley & Sons, Ltd.

Maximum diversity in sample data is required in order to ensure that results from such are representative of the entire domain. In cases where the generation of data is computationally expensive, such as image synthesis, the number of samples should be kept to a minimum with a higher density in regions of transition with respect to image variation. Our objective is to synthesize a set of hyperspectral images to evaluate the performance of ATRs. We use the Digital Imaging and Remote Sensing Image Generation (DIRSIG) model, an image synthesizing software, to generate the images. The nature of a synthesized image is determined by numerous input parameters to DIRSIG. It is required that the resulting image set be diverse with respect to the degree of difficulty for the ATRs under test. We model each synthesized image as a function of the input parameters to DIRSIG, each parameter being a possible source of variation in the image. We compute a Complexity Measure (CM) for each image that represents the degree of difficulty for an ATR. A gradient based sampling scheme is infeasible to determine the regions of transitions in the CM in the multiparameter space because of the computational cost of synthesizing each image. We thus present a sampling algorithm based on an active walker model, in which the step size is adapted based on the distribution of the CM values from the already synthesized images. We sample a variety of multi-dimensional functions with this algorithm, and confirm the improved reconstruction accuracy from samples obtained using it compared to even and random sampling. When applied to sampling the CM multi-parameter space, our adaptive sampling algorithm produces a more diverse image set with respect to degree of difficulty than the random and even sampling schemes.

We propose a novel, progressive, adaptive sampling method, that efficiently varies the sampling rate in local regions of a function based on the distribution of already collected samples. We show that for many functions, increasing the sampling rate in a region of a function with relatively higher complexity is achieved by the equalization of the histogram of the sampled function values. The sampling scheme thus achieves two purposes that are shown to be equivalent; efficiently adapted sampling rates based on local function complexity, and an improvement in diversity in the sampled function values. We achieve the sampling, progressively, through an active walker model. The sample points are placed based on the location of simulated active walkers whose movement is adapted at each stage to achieve the required histogram equalization. The only requirement by this algorithm is the ability to obtain the value of the function at each sample point; no a priori information on the relative levels of local variation of the function being sampled is required. We illustrate this concept with some examples, and discuss practical applications in the two broad areas of adaptive sampling, and improving diversity in samples.

A method to quantify clutter in hyperspectral infrared (HSI) images in a framework similar to work done on single-band images is presented. Hereby, all objects in a scene that may be mistaken for targets by an automatic target recognition (ATR) algorithm are considered clutter. A hyperspectral image contains a number of contiguous discrete bands within the spectrum. The aim is to obtain a measure of complexity for hyperspectral images, which will indicate the inherent difficulty for an ATR to detect targets. We implemented 129 different image clutter metrics, and computed them for a database of synthesized HSI images. A matched filter ATR was used to determine the amount of clutter in the images as a baseline. We developed a method to select a subset of the metrics that in combination correlated best with the amount of clutter in an image, and defined this as the clutter complexity measure (CCM). Multiple runs of this selection procedure for different training image groups show a dominance of a further subset of metrics that best predict the CCM. Our results also show that the CCM obtained from a varying number of random sample images generalizes well for the entire database

Starting with a short review of the available literature in the field of pedestrian and evacuation research, an overview is given over the observed collective phenomena in pedestrian crowds. This includes lane formation in corridors and oscillations at bottlenecks in normal situations, while different kinds of blocked states are produced in panic situations. By means of molecular-dynamic-like microsimulations based on a generalized force model of interactive pedestrian dynamics, the spatio-temporal patterns in pedestrian crowds are successfully reproduced and interpreted as self-organized phenomena. In contrast to previous socio-psychological approaches, this allows a physical understanding of the observations. Despite the significantly different phenomena occuring in normal and panic situations, the main effects can be described by a unified model containing only well interpretable and plausible terms. The transition between the "rational" normal behavior and the apparently "irrational" panic behavior is controlled by a single parameter, the "nervousness", which influences fluctuation strengths, desired speeds, and the tendency of herding. Thereby, it causes paradoxial effects like "freezing by heating", "faster is slower", and the ignorance of available exits. Nevertheless, there are measures to improve pedestrian flows, both in normal and panic situations. For example, the suitable placement of columns can help, although they reduce the accessible space.

Although pedestrians have individual preferences, aims, and destinations, the dynamics of pedestrian crowds is surprisingly predictable. Pedestrians can move freely only at small pedestrian densities. Otherwise their motion is affected by repulsive interactions with other pedestrians, giving rise to self-organization phenomena. Examples of the resulting patterns of motion are separate lanes of uniform walking direction in crowds of oppositely moving pedestrians or oscillations of the passing direction at bottlenecks. If pedestrians leave footprints on deformable ground (for example, in green spaces such as public parks) this additionally causes attractive interactions which are mediated by modifications of their environment. In such cases, systems of pedestrian trails will evolve over time. The corresponding computer simulations are a valuable tool for developing optimized pedestrian facilities and way systems.

This paper extends our development of acoustical bearings-only target localization for the case of multiple moving targets. The resulting techniques can be used to locate and track targets traveling through a network of acoustical sensor arrays. Each array computes and transmits multiple, direction-of-arrival (DOA) estimates to a central processor, which employs the target localization technique. In previous work, we developed ML techniques that may or may not account for the fact that a bearing measurement points to the location of a moving target at a retarded time. By inserting a simple bearings association computation in the ML methods., we define quasi-ML techniques that can estimate the location and velocity of multiple targets using multiple bearing estimates per a sensor array.

Self-organized and error-resistant control of distributed autonomous robotic units in a manufacturing environment with obstacles where the robotic units have to be assigned to manufacturing targets in a cost effective way, is achieved by using two fundamental principles of nature. First, the selection behavior of modes is used which appears in pattern formation of physical, chemical and biological systems. Coupled selection equations based on these pattern formation principles can be used as dynamical system approach to assignment problems. These differential equations guarantee feasibility of the obtained solutions which is of great importance in industrial applications. Second, a model of behavioral forces is used, which has been successfully applied to describe self-organized crowd behavior of pedestrians. This novel approach includes collision avoidance as well as error resistivity. In particular, in systems where failures are of concern, the suggested approach outperforms conventional methods in covering up for sudden external changes like breakdowns of some robotic units. The capability of this system is demonstrated in computer simulations.

We develop four maximum likelihood (ML) methods to localize a moving target using a network of acoustical sensor arrays. Each array transmits a direction-of-arrival (DOA) estimate to a central processor, which employs one of the localization techniques. The four ML approaches use different target signal models where the time retardation factor for the target position and the degradation of the target signal through the air may or may not be included in the model. We compare these methods along with a linear least squares approach through a number of simulations at various signal to noise levels.

The primary purpose of this project(1) is to collect data required to adjust the parameters of a traffic model, allowing realistic predictions of the traffic in Atlanta. This project will provide information with a completeness and accuracy which has not been collected elsewhere. The acquired data will serve the traffic simulation as well as providing an empirical basis for future development in traffic modeling. The verified and calibrated traffic flow model will be implemented in a computer simulation program, which then can be used to focus on the development of strategies for improvement of the traffic flow, optimal positioning for driver-information displays, and evaluation of the significance of the drivers' response rate for the success of control measures.

A new paradigm in ground surveillance consists of swarms of autonomous internetted sensors that can be used for target localization and environmental monitoring. The individual component is an inexpensive device containing multiple sensor types, a processor and wireless communication hardware.Scattered over a certain region, these devices are able to detect the direction or proximity of targets. One of the most limiting factor of the devices is the battery supply. In order to conserve power, these units should be able to adjust their activities to the current situations. Energy consuming signal processing should only be performed if the quality of the raw sensor data promises a significant improvement to the localization results.We propose a self-organized control system that allows the devices to select the algorithm complexity which balances the requirements for good localization performance and energy conservation. The devices make their selection autonomously, based on their own sensor data, information that they receive from other devices in the region, and the amount of energy they have left. The capability of this system will be demonstrated via computer simulations.

Distributed autonomous robotic units are controlled in a manufacturing environment with obstacles where the robotic units have to be assigned to manufacturing targets in a cost effective way. Besides the dynamic control of the robotic units in space, the underlying optimization problem is the NP-hard three-index assignment problem. The suggested approach leads to a self-organized behaviour of the robotic units. The used dynamical system is based first on the selection of modes which appears in pattern formation of physical systems and second on behavioural forces, which have been successfully applied to describe self-organized crowd behaviour of pedestrians. This novel approach includes collision avoidance as well as error resistance, i.e., in systems where failures are of concern, the system covers up for sudden external changes like breakdowns of some robotic units. The capabilities are demonstrated in computer simulations.

A real robotic experiment is done to verify a self-organized control mechanism proposed by J. Starke and P. Molnar for distributed autonomous robotic systems. The problem considered here is the dynamic assignment of two different types of autonomous mobile robotic units to targets in a working area with obstacles. The underlying optimization task is the so-called three-index assignment problem which is NP-hard. Each target has to be served by exactly one robotic unit of each type. While they move to the targets they have to avoid collisions with obstacles and the other robotic units. As soon as they have reached the target they have to perform some tasks cooperatively. This scenario serves as simple model for manufacturing processes.

The task of assigning a team of mobile robotic systems to individual job-locations has many challenges. We use the dynamical systems approach of coupled selection equations to achieve this problem. This intrinsically distributed algorithm has several advantages over traditional integer programs and other distributed approaches: 1) no backtracking is needed, 2) it can be used for NP-hard problems, such as assigning multiple robots with different capabilities to a certain job (three- or higher-index assignment problems), and 3) feasibility of the obtained solutions can be guaranteed.The key point of the applicability in real distributed environments is a distinctive communication fault tolerance so that the necessary data communication between the different processes does not alter to an Achilles heel of the system. Therefore, the present paper addresses the loss of messages in a distributed robotic system based on coupled selection equations, and demonstrates the remarkable communication fault tolerance of this specific dynamical system approach by computer simulations.

The self-organized and fault tolerant behavior of a novel control method for the dynamic assignment of robots to targets using an approach proposed by Starke and Molnar is investigated in detail. Concerning the robot-target assignments the method shows an excellent error resistivity and robustness by using only the local information of each robot. Experimental results verify the dynamic assignment of the mobile robots to the targets and the capability to cope with sudden changes like a breakdown of one of the robots. The dependence of the assignment on the speed of the target-selection dynamics is shown by both experiments and numerical simulations. The results suggest the existence of an optimal value for the speed of the target-selection dynamics.

Concepts of self-organization, adapted from physics, chemistry or biology [4], [3], [7] will become more and more important for the implementation of suitable control mechanisms in the field of artificial intelligent systems and especially in the field of distributed autonomous mobile robotic systems. These concepts seem to be very promising to guarantee the required flexibility, robustness and fault-tolerance.

Pedestrian crowds can very realistically be simulated with a social force model which describes the different influences affecting individual pedestrian motion by a few simple force terms. The model is able to reproduce the emergence of several empirically observed collective patterns of motion. These self-organization phenomena can be utilized for new flow optimization methods which are indispensable for skilful town- and traffic-planning.

A simulation model for the dynamic behaviour of pedestrian crowds is mathematically formulated in terms of a social force model, that means, pedestrians behave in a way as if they would be subject to an acceleration force and to repulsive forces describing the reaction to borders and other pedestrians. The computational simulations presented yield many realistic results that can be compared with video films of pedestrian crowds. Especially, they show the self-organization of collective behavioural patterns. By assuming that pedestrians tend to choose routes that are frequently taken the above model can be extended to an active walker model of trail formation. The topological structure of the evolving trail network will depend on the disadvantage of building new trails and the durability of existing trails. Computer simulations of trail formation indicate to be a valuable tool for designing systems of ways which satisfy the needs of pedestrians best. An example is given for a non-directed trail network.